CN117766133A - Intelligent algorithm-based traditional Chinese medicine syndrome identification method and device - Google Patents
Intelligent algorithm-based traditional Chinese medicine syndrome identification method and device Download PDFInfo
- Publication number
- CN117766133A CN117766133A CN202410024978.5A CN202410024978A CN117766133A CN 117766133 A CN117766133 A CN 117766133A CN 202410024978 A CN202410024978 A CN 202410024978A CN 117766133 A CN117766133 A CN 117766133A
- Authority
- CN
- China
- Prior art keywords
- syndrome
- chinese medicine
- traditional chinese
- model
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 208000011580 syndromic disease Diseases 0.000 title claims abstract description 447
- 239000003814 drug Substances 0.000 title claims abstract description 221
- 238000000034 method Methods 0.000 title claims abstract description 82
- 238000004422 calculation algorithm Methods 0.000 title claims abstract description 29
- 230000004069 differentiation Effects 0.000 claims abstract description 98
- 238000011156 evaluation Methods 0.000 claims abstract description 81
- 238000012360 testing method Methods 0.000 claims abstract description 63
- 238000012549 training Methods 0.000 claims abstract description 29
- 208000024891 symptom Diseases 0.000 claims abstract description 28
- 238000007781 pre-processing Methods 0.000 claims abstract description 21
- 238000013075 data extraction Methods 0.000 claims description 50
- 238000004458 analytical method Methods 0.000 claims description 45
- 238000003745 diagnosis Methods 0.000 claims description 29
- 238000000605 extraction Methods 0.000 claims description 19
- 239000011159 matrix material Substances 0.000 claims description 18
- 238000012937 correction Methods 0.000 claims description 15
- 238000013528 artificial neural network Methods 0.000 claims description 11
- 238000010276 construction Methods 0.000 claims description 11
- 238000010801 machine learning Methods 0.000 claims description 11
- 238000013210 evaluation model Methods 0.000 claims description 9
- 238000007619 statistical method Methods 0.000 claims description 8
- 238000013527 convolutional neural network Methods 0.000 claims description 7
- 230000002457 bidirectional effect Effects 0.000 claims description 6
- 238000004140 cleaning Methods 0.000 claims description 6
- 238000005516 engineering process Methods 0.000 claims description 6
- 238000002372 labelling Methods 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 5
- 238000012216 screening Methods 0.000 claims description 4
- 125000004122 cyclic group Chemical group 0.000 claims description 3
- 238000003062 neural network model Methods 0.000 claims description 3
- 201000010099 disease Diseases 0.000 description 11
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 11
- 238000010586 diagram Methods 0.000 description 8
- 238000004590 computer program Methods 0.000 description 7
- 208000037386 Typhoid Diseases 0.000 description 5
- 238000013461 design Methods 0.000 description 5
- 230000003993 interaction Effects 0.000 description 5
- 238000003860 storage Methods 0.000 description 5
- 201000008297 typhoid fever Diseases 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 230000010354 integration Effects 0.000 description 4
- 239000000463 material Substances 0.000 description 4
- 238000012986 modification Methods 0.000 description 4
- 230000004048 modification Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 230000004075 alteration Effects 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 2
- 230000014509 gene expression Effects 0.000 description 2
- 229940126532 prescription medicine Drugs 0.000 description 2
- 238000004393 prognosis Methods 0.000 description 2
- 230000000306 recurrent effect Effects 0.000 description 2
- 206010043946 Tongue conditions Diseases 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000003930 cognitive ability Effects 0.000 description 1
- 238000010219 correlation analysis Methods 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 229940079593 drug Drugs 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000008506 pathogenesis Effects 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Landscapes
- Medical Treatment And Welfare Office Work (AREA)
Abstract
The invention discloses a traditional Chinese medicine syndrome identification method and device based on an intelligent algorithm. The traditional Chinese medicine syndrome identification method based on the intelligent algorithm comprises the following steps: the method comprises the steps of preprocessing traditional Chinese medicine syndrome text data, and dividing the preprocessed traditional Chinese medicine syndrome text data into syndrome training set data and syndrome test set data according to a certain proportion; based on the syndrome training set data, constructing a syndrome differentiation model; based on the syndrome test set data, testing the syndrome differentiation model to determine the syndrome differentiation model of the traditional Chinese medicine; analyzing the syndrome identification capability of the traditional Chinese medicine syndrome differentiation model to obtain a syndrome identification capability comprehensive evaluation index; and establishing an intelligent auxiliary dialectical interface, and outputting a symptom prediction result through the intelligent auxiliary dialectical interface. According to the method, the syndrome differentiation model is automatically obtained through the automatic learning framework, a plurality of interfaces are established, the data crossing application capability of the syndrome differentiation model is improved, and the problem that the data crossing application capability of the current model is low is solved.
Description
Technical Field
The invention relates to the technical field of computers, in particular to a traditional Chinese medicine syndrome identification method and device based on an intelligent algorithm.
Background
The identification method of the traditional Chinese medicine syndrome is one of the core contents of the traditional Chinese medicine diagnosis, and aims to accurately judge the traditional Chinese medicine syndrome of the disease according to the information of the symptoms, pulse condition, tongue condition and the like of the patient. However, the complexity and subjectivity of traditional Chinese medicine make standardization and improvement of accuracy challenges due to the comprehensive consideration of several factors required for the differentiation of traditional Chinese medicine. Traditional Chinese medicine diagnosis often depends on experience and knowledge of doctors, which results in differentiation of syndrome differentiation results among different doctors. Therefore, researchers are focusing on the use of intelligent algorithms, such as artificial intelligence and machine learning, to improve the objectivity, accuracy and consistency of the identification of the symptoms of traditional Chinese medicine, thereby promoting the modernization and scientization of the diagnosis and treatment of traditional Chinese medicine.
In the modern traditional Chinese medicine syndrome identification method, the application of an intelligent algorithm becomes a leading edge trend. Machine learning and artificial intelligence techniques can extract patterns and rules from a large amount of patient data and traditional Chinese medicine literature by analyzing them to assist doctors in making decisions on the symptoms of traditional Chinese medicine. Among them, deep learning-based methods, such as Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN), have achieved remarkable results in the recognition of the symptoms of traditional Chinese medicine. The algorithms can learn and understand the multi-modal information of the patient, including textual descriptions, images (e.g., tongue), and pulse data. In addition, there are methods for model training that combine expert knowledge with big data.
For example, publication No.: the invention patent of CN113990493A discloses a man-machine interaction intelligent system for diagnosing Chinese medical diseases and identifying symptoms, which comprises the following components: the system comprises an information collection unit for man-machine interaction, a time-sharing description unit for collected information and a diagnosis decision unit, wherein: the man-machine interaction information collection unit organizes the image information related to disease diagnosis/syndrome identification into a multi-level tree structure according to the subordinate relationship among the images; the time-sharing description unit for the collected information is used for describing and displaying the appearance information collected by the man-machine interaction information collection unit in sequence in a two-dimensional form according to the time sequence of appearance information; the diagnosis decision unit comprises diagnosis decision results which are obtained according to the information in the two-dimensional table of the time-sharing description unit of the collected information and are divided by time, and the diagnosis decision results at each time comprise: any diagnosis result is not established, one syndrome/disease is established or a group of syndromes/diseases is established.
For example, bulletin numbers: identification method, system, electronic device and readable storage medium of Chinese medicine certification of CN111768842A, comprising: extracting Chinese medicine identification data from the historical cases, wherein the Chinese medicine identification data comprises a plurality of evidence elements and a plurality of syndromes corresponding to each evidence element; quantifying the syndrome degree of a plurality of syndromes corresponding to the target syndrome factors in each historical case to generate a relation matrix for any target syndrome factors; calculating a relation matrix based on a gray correlation analysis method to obtain gray correlation coefficients of the target certification element and each corresponding certification; calculating the gray correlation coefficient based on a fuzzy information entropy algorithm to obtain the information entropy of the target evidence and each corresponding evidence; and taking the information entropy as the association degree of the target evidence and each corresponding evidence.
However, in the process of implementing the technical scheme of the invention in the embodiment of the application, the application finds that the above technology has at least the following technical problems:
in the prior art, medical record data of different hospitals have heterogeneity, and the problem of low data application crossing capacity of a model exists.
Disclosure of Invention
The embodiment of the application solves the problem of low data application crossing capacity of the model in the prior art by providing the traditional Chinese medicine syndrome identification method and device based on the intelligent algorithm, and improves the data application crossing capacity of the model.
The embodiment of the application provides a traditional Chinese medicine syndrome identification method based on an intelligent algorithm, which comprises the following steps: the method comprises the steps of preprocessing traditional Chinese medicine syndrome text data, and dividing the preprocessed traditional Chinese medicine syndrome text data into syndrome training set data and syndrome test set data according to a certain proportion; based on the syndrome training set data, constructing a syndrome differentiation model; based on the syndrome test set data, testing the syndrome differentiation model to determine the syndrome differentiation model of the traditional Chinese medicine; analyzing the syndrome identification capability of the traditional Chinese medicine syndrome differentiation model to obtain a syndrome identification capability comprehensive evaluation index, wherein the syndrome identification capability comprehensive evaluation index is used for quantifying the syndrome identification capability of the traditional Chinese medicine syndrome differentiation model; and establishing an intelligent auxiliary dialectical interface, and outputting a symptom prediction result through the intelligent auxiliary dialectical interface.
Further, the method for preprocessing the text data of the traditional Chinese medicine syndrome comprises the following steps: data cleaning is carried out on traditional Chinese medicine syndrome text data, wherein the traditional Chinese medicine syndrome text data comprises natural language text symptoms and basic demographic characteristic data of a patient; extracting symptom term text strings from the Chinese medicine symptom text data subjected to data cleaning by using a Chinese medicine data extraction model; text vectorization is carried out on the character strings of the symptomatic terms, the character strings are converted into word vector matrixes, feature screening is carried out, and long tail features are removed; calculating a weighted word vector matrix by using TF-IDF coding, and fusing the weighted word vector matrix with basic demographic characteristic data of a patient on a patient case-by-case level to obtain a two-dimensional matrix form of fused data, wherein the fused data in the two-dimensional matrix form is used for constructing a syndrome differentiation model; based on expert priori knowledge, decomposing the syndrome diagnosis of the patient to obtain syndrome components of syndrome diagnosis, wherein the syndrome components are in a vector form and serve as prediction targets to construct a syndrome differentiation model; if the positive rate of the syndrome component is smaller than the first threshold index or the number of positive examples is smaller than the second threshold index, the syndrome component is recorded to be unsuitable for modeling prediction, and the syndrome component unsuitable for modeling prediction is removed.
Further, the method for constructing the traditional Chinese medicine data extraction model comprises the following steps: labeling and extracting 'principle prescription' data from the traditional Chinese medicine electronic text by using an electronic text labeling tool, and taking the 'principle prescription' data as modeling data of a traditional Chinese medicine data extraction model; based on modeling data, combining a neural network language model, a bidirectional cyclic neural network model and a convolutional neural network model to construct a plurality of preliminary traditional Chinese medicine data extraction models; repeating modeling of each preliminary traditional Chinese medicine data extraction model for a certain number of times, selecting an evaluation index, and carrying out statistical analysis on modeling data; and determining a preliminary traditional Chinese medicine data extraction model with the optimal evaluation index as a traditional Chinese medicine data extraction model according to the result of the statistical analysis.
Further, the method for constructing the syndrome differentiation model comprises the following steps: training the syndrome training set data by adopting an Autogluon automatic learning frame, and automatically constructing a plurality of machine learning models; and performing integrated learning on the machine learning model based on a Stacking integrated learning technology to obtain a syndrome differentiation model.
Further, the method for determining the syndrome differentiation model of the traditional Chinese medicine syndrome comprises the following steps: predicting the syndrome test set data based on a syndrome differentiation model; and obtaining a syndrome component prediction result and a syndrome component positive rate in the syndrome test set data according to the syndrome differentiation model.
Further, the intelligent auxiliary syndrome differentiation method comprises the following steps: patient information is acquired: preprocessing the input syndrome data; and (3) predicting a loading model: loading a pre-trained traditional Chinese medicine syndrome differentiation model to identify and predict the pre-processed syndrome data, and predicting the syndrome component positive rate of the patient; outputting a result: the positive rate of the syndrome component of the patient is stored and output in the form of a dictionary.
Further, the method for obtaining the syndrome identification capability of the traditional Chinese medicine syndrome differentiation model comprises the following steps: the method comprises the steps of collecting the accuracy of a traditional Chinese medicine syndrome differentiation model on syndrome test set data and the area under an ROC curve, and analyzing the accuracy and the area under the ROC curve to obtain a syndrome identification capability comprehensive evaluation index; when the comprehensive evaluation index of the syndrome identification capability is higher than the threshold value III, the traditional Chinese medicine syndrome identification model is recorded to have excellent syndrome identification capability; when the comprehensive evaluation index of the syndrome identification capability is higher than a threshold value four, recording that the traditional Chinese medicine syndrome identification model has certain syndrome identification capability, wherein the threshold value four is smaller than a threshold value three; when the comprehensive evaluation index of the syndrome identification capability is not higher than the threshold value by four, recording that the traditional Chinese medicine syndrome identification model does not reach the practical level yet; and outputting comprehensive evaluation of the syndrome identification capability of the traditional Chinese medicine syndrome identification model.
Further, the analysis method of the comprehensive evaluation index of the syndrome identification capability comprises the following steps: constructing a syndrome identification capability comprehensive evaluation model according to the accuracy of the traditional Chinese medicine syndrome identification model on syndrome test set data and the area under the ROC curve; calculating a comprehensive evaluation index of the syndrome identification capability according to the comprehensive evaluation model of the syndrome identification capability; the comprehensive evaluation model of the syndrome identification capability is as follows:in (1) the->For the comprehensive evaluation index of the syndrome identification capability corresponding to the syndrome data in the syndrome test set data, KA is an accuracy influence coefficient, gamma 1 is the weight ratio of the accuracy influence coefficient in the comprehensive evaluation index of the syndrome identification capability, KR is the curve area influence coefficient of the traditional Chinese medicine syndrome differentiation model on the syndrome test set dataGamma 2 is the weight ratio of the curve area influence coefficient in the comprehensive evaluation index of the syndrome identification capability, KJ is the syndrome positive influence coefficient, gamma 3 is the weight ratio of the syndrome positive influence coefficient in the comprehensive evaluation index of the syndrome identification capability, KT is the extraction accuracy of the traditional Chinese medicine data, gamma 4 is the weight ratio of the extraction accuracy of the traditional Chinese medicine data in the comprehensive evaluation index of the syndrome identification capability, delta is the correction factor of the comprehensive evaluation index of the syndrome identification capability, and e is a natural constant; the method for acquiring the accuracy influence coefficient comprises the following steps: constructing an accuracy influence coefficient analysis model according to the accuracy of the traditional Chinese medicine syndrome differentiation model on the syndrome test set data; calculating an accuracy influence coefficient according to the accuracy influence coefficient analysis model; the accuracy influence coefficient analysis model is as follows: / >Wherein ACC0 is an accuracy threshold, ACC is accuracy, < >>A correction factor for the accuracy influencing coefficient; the method for acquiring the curve area influence coefficient comprises the following steps: constructing a curve area influence coefficient analysis model according to the area under the ROC curve of the traditional Chinese medicine syndrome differentiation model on the syndrome test set data; calculating a curve area influence coefficient according to the curve area influence coefficient analysis model; the curve area influence coefficient analysis model is as follows: />Wherein AUC0 is the area under ROC curve threshold, AUC is the area under ROC curve,/->A correction factor for the curve area influence coefficient; the analysis method of the syndrome positive influence coefficient comprises the following steps: obtaining the syndrome positive rate and the syndrome positive number of the traditional Chinese medicine syndrome differentiation model on the syndrome test set data; constructing a syndrome positive influence coefficient analysis model according to the syndrome positive rate and the syndrome positive number; analysis model based on syndrome positive influence coefficientCalculating positive influence coefficients of syndrome; the syndrome positive influence coefficient analysis model is as follows:wherein, theta 0 is a syndrome positive rate threshold value, theta is a syndrome positive rate, delta theta is a set unit value of the syndrome positive rate, kappa 1 is a weight ratio of the syndrome positive rate in a syndrome positive influence coefficient, L0 is a syndrome positive instance number threshold value, L is a syndrome positive instance number, delta L is a set unit value of the syndrome positive instance number, kappa 2 is a weight ratio of the syndrome positive instance number in the syndrome positive influence coefficient >Is a correction factor for syndrome positive influence coefficient.
Further, the analysis method of the extraction accuracy of the traditional Chinese medicine data comprises the following steps: acquiring evaluation indexes of a traditional Chinese medicine data extraction model, wherein the evaluation indexes comprise F1 scores; constructing a syndrome Chinese medicine data extraction accuracy analysis model according to the evaluation indexes; calculating the extraction accuracy of the traditional Chinese medicine data according to the analysis model of the extraction accuracy of the traditional Chinese medicine data; the traditional Chinese medicine data extraction accuracy analysis model is as follows:wherein F is an evaluation index, F0 is an evaluation index threshold, deltaF is an evaluation index set unit value, & lt/EN & gt>And extracting correction factors of accuracy rate for the traditional Chinese medicine data.
The embodiment of the application provides a traditional Chinese medicine syndrome identification device based on an intelligent algorithm, which comprises a data preprocessing module, a syndrome differentiation model construction module, a syndrome differentiation model test module, a syndrome identification capability assessment module and an auxiliary syndrome differentiation prediction module: the data preprocessing module is used for: the method comprises the steps of preprocessing traditional Chinese medicine syndrome text data, and dividing the preprocessed traditional Chinese medicine syndrome text data into syndrome training set data and syndrome test set data according to a certain proportion; the dialectical model construction module is used for: the method is used for constructing a syndrome differentiation model based on syndrome training set data; the dialectical model test module is as follows: the method is used for testing the syndrome differentiation model based on the syndrome test set data to determine the traditional Chinese medicine syndrome differentiation model; the syndrome identification capability evaluation module: the comprehensive evaluation index of the syndrome identification capability is used for quantifying the syndrome identification capability of the traditional Chinese medicine syndrome differentiation model; the auxiliary dialectical prediction module is used for: the intelligent diagnosis interface is used for establishing an intelligent auxiliary diagnosis interface, and the prediction result of the syndrome is output through the intelligent auxiliary diagnosis interface.
One or more technical solutions provided in the embodiments of the present application at least have the following technical effects or advantages:
1. the traditional Chinese medicine data extraction model is set, and the syndrome differentiation model with better performance is automatically obtained by utilizing the automatic learning frame, so that corresponding network calling interfaces are set for each model, the data crossing application capacity of the model is improved, and the problem that the data crossing application capacity of the model is lower in the prior art is effectively solved.
2. The network call interface of the traditional Chinese medicine 'principle medicine' data extraction model is established in the form of the model call of the network service interface through the accessibility and convenience of the traditional Chinese medicine data extraction model from the use and the integration angle fused to other applications, so that the accessibility is high, the use is convenient, a user does not need to add required hardware support, and the economic cost is saved.
3. The intelligent auxiliary dialectical interface is established, and the result of the prediction of the syndrome is output through the intelligent auxiliary dialectical interface, so that complex program files are not required to be downloaded, complicated operation system configuration and environment construction work are performed, and further time cost is saved.
Drawings
Fig. 1 is a flowchart of a method for identifying traditional Chinese medicine syndrome based on an intelligent algorithm according to an embodiment of the present application;
fig. 2 is a block diagram of a traditional Chinese medicine syndrome identification device based on an intelligent algorithm according to an embodiment of the present application.
Detailed Description
According to the method and the device for identifying the traditional Chinese medicine syndrome based on the intelligent algorithm, the problem that the data crossing application capacity of a model is low in the prior art is solved, the syndrome differentiation model is automatically obtained through an automatic learning framework, a network call interface is set for the model, and the data crossing application capacity of the model is improved.
The technical scheme in the embodiment of the application aims to solve the problem that the data-crossing application capability of the existing model is low, and the overall thought is as follows:
the method comprises the steps of obtaining Chinese medicine syndrome text data, preprocessing the Chinese medicine syndrome text data, and dividing the Chinese medicine syndrome text data into a syndrome training set and a test set according to a certain proportion. And constructing a syndrome differentiation model by using the syndrome training set, and then evaluating the model performance by using the test set to determine the traditional Chinese medicine syndrome differentiation model. And analyzing the syndrome identification capability of the model to obtain a comprehensive evaluation index, and quantifying the syndrome identification capability of the model. And finally, establishing an intelligent auxiliary syndrome differentiation interface, and outputting a prediction result of the traditional Chinese medicine syndrome through the interface, so that the data crossing application capability of the model is improved.
In order to better understand the above technical solutions, the following detailed description will refer to the accompanying drawings and specific embodiments.
As shown in fig. 1, a flowchart of a method for identifying traditional Chinese medicine syndrome based on an intelligent algorithm according to an embodiment of the present application is provided, and the method is applied to a device for identifying traditional Chinese medicine syndrome based on an intelligent algorithm, and includes the following steps: the method comprises the steps of preprocessing traditional Chinese medicine syndrome text data, and dividing the preprocessed traditional Chinese medicine syndrome text data into syndrome training set data and syndrome test set data according to a certain proportion; based on the syndrome training set data, constructing a syndrome differentiation model; based on the syndrome test set data, testing the syndrome differentiation model to determine the syndrome differentiation model of the traditional Chinese medicine; analyzing the syndrome identification capability of the traditional Chinese medicine syndrome differentiation model to obtain a syndrome identification capability comprehensive evaluation index, wherein the syndrome identification capability comprehensive evaluation index is used for quantifying the syndrome identification capability of the traditional Chinese medicine syndrome differentiation model; and establishing an intelligent auxiliary dialectical interface, and outputting a symptom prediction result through the intelligent auxiliary dialectical interface.
In this embodiment, the "theory of medicine" structure system of "typhoid treatises" is used as guidance to collect modeling data, and the named entity recognition (Named entity recognition, NER) modeling method and the bidirectional coding representation (Bidirectional encoder representations from transformer, BERT) neural network language model based on the transducer are used as reference to obtain the modeling data collection scheme of the traditional Chinese medicine "theory medicine" data extraction model and the construction method of the extraction model, and the research result is transformed from the aspects of accessibility, convenience and integration, and the application support is provided for the automatic extraction of traditional Chinese medicine data through the calling form of the network interface.
Further, the method for preprocessing the traditional Chinese medicine syndrome text data comprises the following steps: data cleaning is carried out on traditional Chinese medicine syndrome text data, wherein the traditional Chinese medicine syndrome text data comprises natural language text symptoms and basic demographic characteristic data of a patient; extracting symptom term text strings from the Chinese medicine symptom text data subjected to data cleaning by using a Chinese medicine data extraction model; text vectorization is carried out on the character strings of the symptomatic terms, the character strings are converted into word vector matrixes, feature screening is carried out, and long tail features are removed; calculating a weighted word vector matrix by using TF-IDF codes, and fusing the weighted word vector matrix with basic demographic characteristic data of a patient on a patient case-by-case level to obtain a two-dimensional matrix form of fused data, wherein the fused data in the two-dimensional matrix form is used for constructing a syndrome differentiation model; based on expert priori knowledge, decomposing the syndrome diagnosis of the patient to obtain syndrome components of the syndrome diagnosis, wherein the syndrome components are in a vector form and serve as prediction targets to construct a syndrome differentiation model; if the positive rate of the syndrome component is smaller than the first threshold index or the number of positive examples is smaller than the second threshold index, the syndrome component is recorded to be unsuitable for modeling prediction, and the syndrome component unsuitable for modeling prediction is removed.
In this embodiment, for certain disease type syndrome data, all available data of the disease type syndrome data are randomly sampled in layers according to syndrome categories, 70% is used as syndrome training set data, and 30% is used as syndrome test set data. The syndrome test set data is not involved in any stage of model training and is only used for estimating the generalization performance of the model after model training is completed. The model input data is multimodal data, and consists of a terminology symptom text string (text mode) and basic demographic characteristic data (matrix mode) of a patient, wherein the terminology symptom text string (text mode) is output by the natural language text symptom extraction module. Model training is carried out in a multi-mode pre-fusion mode: firstly, a text vectorization technology is utilized to convert symptom text strings into word vector matrixes, and then long tail features (the cumulative frequency is 95-100%) are counted and screened out. After feature screening, word frequency-inverse text frequency (term frequency-inverse document frequency, tf-idf) coding is performed, and a weighted word vector matrix is calculated. And fusing the weighted word vector matrix and the demographic characteristic data on the individual case level of the patient, thereby obtaining a two-dimensional matrix form of the training data. On the other hand, the patient syndrome diagnosis is decomposed based on expert priori knowledge to obtain syndrome components (vector forms) of syndrome diagnosis, the model is trained as a prediction target, and if the positive rate of the syndrome components is less than 1% or the number of positive examples is less than 10, the syndrome components are recorded to be unsuitable for modeling prediction of the syndrome differentiation model.
Further, the method for constructing the traditional Chinese medicine data extraction model comprises the following steps: labeling and extracting 'principle prescription' data from the traditional Chinese medicine electronic text by using an electronic text labeling tool, and taking the 'principle prescription' data as modeling data of a traditional Chinese medicine data extraction model; based on modeling data, combining a neural network language model, a bidirectional cyclic neural network model and a convolutional neural network model to construct a plurality of preliminary traditional Chinese medicine data extraction models; repeating modeling of each preliminary traditional Chinese medicine data extraction model for a certain number of times, selecting an evaluation index, and carrying out statistical analysis on modeling data; and determining a preliminary traditional Chinese medicine data extraction model with the optimal evaluation index as a traditional Chinese medicine data extraction model according to the result of the statistical analysis.
In this embodiment, the teaching materials of traditional Chinese medicine, the clinical electronic medical records of traditional Chinese medicine and famous doctors which are rich in "medicine of theory" data are selected as raw data sources, characters belonging to "medicine of theory" and corresponding categories in the electronic text are marked by using an electronic text marking tool, the characters are used as modeling data of an intelligent extraction model, data support is provided for establishing the data extraction model of "medicine of theory" of traditional Chinese medicine, and the electronic text marking tool comprises a brat electronic text marking system. The traditional Chinese medicine electronic text is taken as modeling data and comprises electronic texts of three teaching materials of typhoid theory lecture, traditional Chinese medicine internal science and prescription science. The treatise on typhoid is the ancestor of prescription, the meridian of medical prescription is the monograph of syndrome differentiation and the contents of the theory of typhoid are read from the aspect of teaching, and include a plurality of classical "rational prescription" knowledge expressions. The Chinese medicine science and the prescription science are the teaching materials close to the clinical practice of the current Chinese medicine, and the knowledge of the principle prescription medicine is also covered, so that the contents of the diseases, symptoms, diseases, etiology, pathogenesis, treatment methods, prescriptions, medicines, efficacy, dosage and the like of the Chinese medicine are comprehensively related in each section of the teaching materials. Therefore, three books of the theory of typhoid fever lecture, chinese medicine science, prescription science, can be used as modeling data to enable a model to simulate the study of basic theoretical knowledge of Chinese medicine, and endow the model with the cognitive ability of 'principle prescription' to the theory of Chinese medicine. The traditional Chinese medicine 'theory medicine' data is extracted from continuous characters, and the modeling design of a 'theory medicine' data extraction model is carried out from NER model modeling thought good at extracting specific data from character sequences, which mainly relates to the following steps: 1. model design based on Bi-cyclic neural network (Bidirectional recurrent neural network, bi-RNN) structure; 2. model design based on the combination of Bi-RNN and convolutional neural network (Convolution neural network, CNN); 3. model design taking the BERT neural network language model as a main body and model design based on the combination of the BERT neural network language model and the Bi-RNN. And (3) taking the accuracy rate, the recall rate and the F1-score as evaluation indexes, repeating modeling for 10 times for each model, analyzing by using a statistical analysis method, and determining the model with the best effect and the contribution of modeling data to the model according to the result of the statistical analysis. Based on NER model modeling thought, a plurality of modeling schemes of traditional Chinese medicine 'theory prescription' data extraction models are provided, modeling tests are carried out, and performances of different models are compared. According to the test result, a traditional Chinese medicine 'theory prescription' data extraction model with the best effect, namely a traditional Chinese medicine data extraction model, is obtained. The modeling data beneficial to the model expression is utilized to construct a preliminary traditional Chinese medicine data extraction model capable of extracting 10 traditional Chinese medicine 'principle prescription' data, and a network call interface of the model is constructed, so that an intelligent tool is provided for high-efficiency automatic extraction of the traditional Chinese medicine 'principle prescription' data. In order to realize the construction of a traditional Chinese medicine 'principle prescription' data extraction model, an intelligent model for application is formed, and a modeling data acquisition scheme of the traditional Chinese medicine 'principle prescription' data extraction model is formulated; the construction scheme of the traditional Chinese medicine 'theory prescription' data extraction model; and application schemes of model test and model. The traditional Chinese medicine data extraction model establishes a network call interface of the traditional Chinese medicine 'principle prescription medicine' data extraction model in a model call form of a network service interface from the aspects of accessibility and convenience of use and integration into other applications. On the one hand, based on the network port, the accessibility is high, the use is convenient, the user does not need to add the required hardware support, and the economic cost is saved; the complex program file is not required to be downloaded, and complicated operation system configuration and environment construction work are carried out, so that the problem that the system cannot be used due to failure in configuration is avoided, and the time cost is saved. On the other hand, the network service interface can be conveniently integrated into different applications, and provides data extraction support for related applications.
Further, the construction method of the syndrome differentiation model comprises the following steps: training the syndrome training set data by adopting an Autogluon automatic learning frame, and automatically constructing a plurality of machine learning models; and performing integrated learning on the machine learning model based on a Stacking integrated learning technology to obtain a syndrome differentiation model.
In this example, the model training process was performed in a Python environment using autopluon 0.5.2. Autogluon is an automatic machine learning framework based on a Stacking integrated learning technology, and a plurality of machine learning models (weak learners) are automatically built from a plurality of machine learning models such as K-nearest neighbor, a support vector machine, random forest, catBoost, XGBoost, deep neural networks with different structures and the like, and then integrated learning is carried out through Stacking, so that a final model with better performance is obtained.
Further, the method for determining the syndrome differentiation model of the traditional Chinese medicine syndrome comprises the following steps: predicting the syndrome test set data based on a syndrome differentiation model; and obtaining a syndrome component prediction result and a syndrome component positive rate in the syndrome test set data according to the syndrome differentiation model.
In this example, according to the individual diagnosis and prognosis prediction model report specifications (Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis, TRIPOD), a model test was performed using a leave-out method as an internal verification process of the model.
Further, the intelligent auxiliary syndrome differentiation method comprises the following steps: patient information is acquired: preprocessing the input syndrome data; and (3) predicting a loading model: loading a pre-trained traditional Chinese medicine syndrome differentiation model to identify and predict the pre-processed syndrome data, and predicting the syndrome component positive rate of the patient; outputting a result: the positive rate of the syndrome component of the patient is stored and output in the form of a dictionary.
In this embodiment, preprocessing the inputted syndrome data includes extracting therefrom basic information (id, disease, sex, age) of the patient and symptom term text.
Further, the method for acquiring the syndrome identification capability of the traditional Chinese medicine syndrome differentiation model comprises the following steps: the method comprises the steps of collecting the accuracy of a traditional Chinese medicine syndrome differentiation model on syndrome test set data and the area under an ROC curve, and analyzing the accuracy and the area under the ROC curve to obtain a syndrome identification capability comprehensive evaluation index; when the comprehensive evaluation index of the syndrome identification capability is higher than the threshold value III, the traditional Chinese medicine syndrome identification model is recorded to have excellent syndrome identification capability; when the comprehensive evaluation index of the syndrome identification capability is higher than a threshold value four, recording that the traditional Chinese medicine syndrome identification model has certain syndrome identification capability, wherein the threshold value four is smaller than the threshold value three; when the comprehensive evaluation index of the syndrome identification capability is not higher than the threshold value by four, recording that the traditional Chinese medicine syndrome identification model does not reach the practical level yet; and outputting comprehensive evaluation of the syndrome identification capability of the traditional Chinese medicine syndrome identification model.
In this embodiment, the model evaluation uses the accuracy of the model on the test set and the area under the ROC curve (Area Under the Receiver Operating Characteristic Curve, AUC) to perform comprehensive evaluation to estimate the ability of the model to perform syndrome identification on unknown patients.
Further, the analysis method of the comprehensive evaluation index of the syndrome identification capability comprises the following steps: constructing a syndrome identification capability comprehensive evaluation model according to the accuracy of the traditional Chinese medicine syndrome identification model on syndrome test set data and the area under the ROC curve; calculating a comprehensive evaluation index of the syndrome identification capability according to the comprehensive evaluation model of the syndrome identification capability; the syndrome identification capability comprehensive evaluation model is as follows:in (1) the->For the comprehensive evaluation index of the syndrome identification capability corresponding to the syndrome data in the syndrome test set data, KA is an accuracy influence coefficient, gamma 1 is a weight occupation ratio of the accuracy influence coefficient in the comprehensive evaluation index of the syndrome identification capability, KR is a curve area influence coefficient of a syndrome differentiation model of traditional Chinese medicine on the syndrome test set data, gamma 2 is a weight occupation ratio of the curve area influence coefficient in the comprehensive evaluation index of the syndrome identification capability, KJ is a positive influence coefficient of the syndrome, gamma 3 is a weight occupation ratio of the positive influence coefficient of the syndrome in the comprehensive evaluation index of the syndrome identification capability, KT is an extraction accuracy of the traditional Chinese medicine data, gamma 4 is a weight occupation ratio of the extraction accuracy of the traditional Chinese medicine data in the comprehensive evaluation index of the syndrome identification capability, delta is a correction factor of the comprehensive evaluation index of the syndrome identification capability, and e is a natural constant; the method for acquiring the accuracy influence coefficient comprises the following steps: constructing an accuracy influence coefficient analysis model according to the accuracy of the traditional Chinese medicine syndrome differentiation model on the syndrome test set data; calculating an accuracy influence coefficient according to the accuracy influence coefficient analysis model; the accuracy influence coefficient analysis model is as follows: / >Wherein ACC0 is an accuracy threshold, ACC is accuracy, < >>A correction factor for the accuracy influencing coefficient; the method for acquiring the curve area influence coefficient comprises the following steps: constructing a curve area influence coefficient analysis model according to the area under the ROC curve of the traditional Chinese medicine syndrome differentiation model on the syndrome test set data; calculating a curve area influence coefficient according to the curve area influence coefficient analysis model; the curve area influence coefficient analysis model is as follows:wherein AUC0 is the area under ROC curve threshold, AUC is the area under ROC curve,/->A correction factor for the curve area influence coefficient; the analysis method of the syndrome positive influence coefficient comprises the following steps: obtaining the syndrome positive rate and the syndrome positive number of the traditional Chinese medicine syndrome differentiation model on the syndrome test set data; constructing a syndrome positive influence coefficient analysis model according to the syndrome positive rate and the syndrome positive number; according to the syndrome positive influence coefficient analysis model, calculating a syndrome positive influence coefficient; the syndrome positive influence coefficient analysis model is as follows: />Wherein, theta 0 is a syndrome positive rate threshold value, theta is a syndrome positive rate, delta theta is a set unit value of the syndrome positive rate, kappa 1 is a weight ratio of the syndrome positive rate in a syndrome positive influence coefficient, L0 is a syndrome positive instance number threshold value, L is a syndrome positive instance number, delta L is a set unit value of the syndrome positive instance number, kappa 2 is a weight ratio of the syndrome positive instance number in the syndrome positive influence coefficient >Is a correction factor for syndrome positive influence coefficient.
In this example, the syndrome positive rate set to 1% and the syndrome positive number set to 1%. The area under the ROC curve (abbreviated as AUC), the closer the AUC is to 1, the better the performance of the model at different operating points.
Further, the analysis method of the extraction accuracy of the traditional Chinese medicine data comprises the following steps: acquiring evaluation indexes of a traditional Chinese medicine data extraction model, wherein the evaluation indexes comprise F1 scores; constructing a syndrome Chinese medicine data extraction accuracy analysis model according to the evaluation indexes; calculating the extraction accuracy of the traditional Chinese medicine data according to the analysis model of the extraction accuracy of the traditional Chinese medicine data; the traditional Chinese medicine data extraction accuracy analysis model is as follows:wherein F is an evaluation index, F0 is an evaluation index threshold, deltaF is an evaluation index set unit value, & lt/EN & gt>And extracting correction factors of accuracy rate for the traditional Chinese medicine data.
In this embodiment, the evaluation index sets the unit value to 1%. The higher the F1 score is, the better the F1 score is, the more the F1 score is in the range of 0 to 1, the closer the F1 score is to 1, the better balance between the accuracy rate and the recall rate is achieved by the model, namely the better balance between the prediction accuracy of the positive examples and the coverage rate of all the positive examples is achieved.
As shown in fig. 2, for a structure diagram of a traditional Chinese medicine syndrome identification device based on an intelligent algorithm provided in an embodiment of the present application, the traditional Chinese medicine syndrome identification device based on an intelligent algorithm provided in an embodiment of the present application includes: the system comprises a data preprocessing module, a syndrome differentiation model construction module, a syndrome differentiation model test module, a syndrome identification capability assessment module and an auxiliary syndrome differentiation prediction module: and a data preprocessing module: the method comprises the steps of preprocessing traditional Chinese medicine syndrome text data, and dividing the preprocessed traditional Chinese medicine syndrome text data into syndrome training set data and syndrome test set data according to a certain proportion; the dialectical model building module: the method is used for constructing a syndrome differentiation model based on syndrome training set data; the dialectical model test module: the method is used for testing the syndrome differentiation model based on the syndrome test set data to determine the traditional Chinese medicine syndrome differentiation model; the syndrome identification capability evaluation module: the comprehensive evaluation index is used for quantifying the syndrome recognition capability of the traditional Chinese medicine syndrome differentiation model; the auxiliary dialectical prediction module: the intelligent diagnosis interface is used for establishing an intelligent auxiliary diagnosis interface, and the prediction result of the syndrome is output through the intelligent auxiliary diagnosis interface.
The technical scheme in the embodiment of the application at least has the following technical effects or advantages: relative to publication No.: according to the man-machine interaction intelligent system for traditional Chinese medicine disease diagnosis and syndrome identification disclosed in the patent of CN113990493A, in the embodiment of the application, a traditional Chinese medicine data extraction model is arranged, and an automatic learning frame is utilized to automatically learn and obtain a syndrome differentiation model with good performance, so that corresponding network calling interfaces are arranged for each model, and the data crossing application capacity of the model is improved; relative to publication No.: the identification method, the identification system, the electronic equipment and the readable storage medium of the traditional Chinese medicine evidence element disclosed by the invention patent publication of CN111768842A are used for establishing a network call interface of a traditional Chinese medicine 'principle medicine' data extraction model in a model call form of a network service interface through accessibility and convenience of use and integration of the traditional Chinese medicine data extraction model into other applications, so that accessibility is high, use is convenient, a user does not need to add required hardware support, and economic cost is saved.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. It is therefore intended that the following claims be interpreted as including the preferred embodiments and all such alterations and modifications as fall within the scope of the invention.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.
Claims (10)
1. The traditional Chinese medicine syndrome identification method based on the intelligent algorithm is characterized by comprising the following steps of:
The method comprises the steps of preprocessing traditional Chinese medicine syndrome text data, and dividing the preprocessed traditional Chinese medicine syndrome text data into syndrome training set data and syndrome test set data according to a certain proportion;
based on the syndrome training set data, constructing a syndrome differentiation model;
based on the syndrome test set data, testing the syndrome differentiation model to determine the syndrome differentiation model of the traditional Chinese medicine;
analyzing the syndrome identification capability of the traditional Chinese medicine syndrome differentiation model to obtain a syndrome identification capability comprehensive evaluation index, wherein the syndrome identification capability comprehensive evaluation index is used for quantifying the syndrome identification capability of the traditional Chinese medicine syndrome differentiation model;
and establishing an intelligent auxiliary dialectical interface, and outputting a symptom prediction result through the intelligent auxiliary dialectical interface.
2. The method for identifying traditional Chinese medicine syndrome based on intelligent algorithm as claimed in claim 1, wherein the method for preprocessing the traditional Chinese medicine syndrome text data is as follows:
data cleaning is carried out on traditional Chinese medicine syndrome text data, wherein the traditional Chinese medicine syndrome text data comprises natural language text symptoms and basic demographic characteristic data of a patient;
extracting symptom term text strings from the Chinese medicine symptom text data subjected to data cleaning by using a Chinese medicine data extraction model;
Text vectorization is carried out on the character strings of the symptomatic terms, the character strings are converted into word vector matrixes, feature screening is carried out, and long tail features are removed;
calculating a weighted word vector matrix by using TF-IDF coding, and fusing the weighted word vector matrix with basic demographic characteristic data of a patient on a patient case-by-case level to obtain a two-dimensional matrix form of fused data, wherein the fused data in the two-dimensional matrix form is used for constructing a syndrome differentiation model;
based on expert priori knowledge, decomposing the syndrome diagnosis of the patient to obtain syndrome components of syndrome diagnosis, wherein the syndrome components are in a vector form and serve as prediction targets to construct a syndrome differentiation model;
if the positive rate of the syndrome component is smaller than the first threshold index or the number of positive examples is smaller than the second threshold index, the syndrome component is recorded to be unsuitable for modeling prediction, and the syndrome component unsuitable for modeling prediction is removed.
3. The method for identifying traditional Chinese medicine symptoms based on intelligent algorithm as claimed in claim 1, wherein the method for constructing the traditional Chinese medicine data extraction model is as follows:
labeling and extracting 'principle prescription' data from the traditional Chinese medicine electronic text by using an electronic text labeling tool, and taking the 'principle prescription' data as modeling data of a traditional Chinese medicine data extraction model;
Based on modeling data, combining a neural network language model, a bidirectional cyclic neural network model and a convolutional neural network model to construct a plurality of preliminary traditional Chinese medicine data extraction models;
repeating modeling of each preliminary traditional Chinese medicine data extraction model for a certain number of times, selecting an evaluation index, and carrying out statistical analysis on modeling data;
and determining a preliminary traditional Chinese medicine data extraction model with the optimal evaluation index as a traditional Chinese medicine data extraction model according to the result of the statistical analysis.
4. The method for identifying traditional Chinese medicine syndrome based on intelligent algorithm as claimed in claim 1, wherein the method for constructing syndrome differentiation model is as follows:
training the syndrome training set data by adopting an Autogluon automatic learning frame, and automatically constructing a plurality of machine learning models;
and performing integrated learning on the machine learning model based on a Stacking integrated learning technology to obtain a syndrome differentiation model.
5. The intelligent algorithm-based traditional Chinese medicine syndrome identification method as claimed in claim 1, wherein the method comprises the following steps: the method for determining the traditional Chinese medicine syndrome differentiation model comprises the following steps:
predicting the syndrome test set data based on a syndrome differentiation model;
and obtaining a syndrome component prediction result and a syndrome component positive rate in the syndrome test set data according to the syndrome differentiation model.
6. The method for identifying traditional Chinese medicine symptoms based on intelligent algorithm as claimed in claim 1, wherein the method for intelligently assisting syndrome differentiation is as follows:
patient information is acquired: preprocessing the input syndrome data;
and (3) predicting a loading model: loading a pre-trained traditional Chinese medicine syndrome differentiation model to identify and predict the pre-processed syndrome data, and predicting the syndrome component positive rate of the patient;
outputting a result: the positive rate of the syndrome component of the patient is stored and output in the form of a dictionary.
7. The method for identifying the traditional Chinese medicine syndrome based on the intelligent algorithm according to claim 1, wherein the method for acquiring the syndrome identification capability of the traditional Chinese medicine syndrome differentiation model is as follows:
the method comprises the steps of collecting the accuracy of a traditional Chinese medicine syndrome differentiation model on syndrome test set data and the area under an ROC curve, and analyzing the accuracy and the area under the ROC curve to obtain a syndrome identification capability comprehensive evaluation index;
when the comprehensive evaluation index of the syndrome identification capability is higher than the threshold value III, the traditional Chinese medicine syndrome identification model is recorded to have excellent syndrome identification capability;
when the comprehensive evaluation index of the syndrome identification capability is higher than a threshold value four, recording that the traditional Chinese medicine syndrome identification model has certain syndrome identification capability, wherein the threshold value four is smaller than a threshold value three;
When the comprehensive evaluation index of the syndrome identification capability is not higher than the threshold value by four, recording that the traditional Chinese medicine syndrome identification model does not reach the practical level yet;
and outputting comprehensive evaluation of the syndrome identification capability of the traditional Chinese medicine syndrome identification model.
8. The method for identifying traditional Chinese medicine symptoms based on intelligent algorithm according to claim 7, wherein the analysis method of the comprehensive evaluation index of the syndrome identification capability is as follows:
constructing a syndrome identification capability comprehensive evaluation model according to the accuracy of the traditional Chinese medicine syndrome identification model on syndrome test set data and the area under the ROC curve;
calculating a comprehensive evaluation index of the syndrome identification capability according to the comprehensive evaluation model of the syndrome identification capability;
the comprehensive evaluation model of the syndrome identification capability is as follows:
in the method, in the process of the invention,for comprehensively evaluating the index for the syndrome identification capability corresponding to the syndrome data in the syndrome test set data, KA is an accuracy influence coefficient, and gamma 1 For the weight ratio of the accuracy influence coefficient in the syndrome identification capability comprehensive evaluation index, KR is the curve area influence coefficient of the traditional Chinese medicine syndrome differentiation model on the syndrome test set data, and gamma 2 The weight ratio of the curve area influence coefficient in the syndrome identification capability comprehensive evaluation index is KJ, the syndrome positive influence coefficient is gamma 3 For the weight ratio of the syndrome positive influence coefficient in the syndrome identification capacity comprehensive evaluation index, KT is the extraction accuracy of traditional Chinese medicine data, and gamma 4 The weight ratio of the traditional Chinese medicine data extraction accuracy rate in the syndrome identification capability comprehensive evaluation index is delta, the syndrome identification capability comprehensive evaluation index correction factor is delta, and e is a natural constant;
the method for acquiring the accuracy influence coefficient comprises the following steps:
constructing an accuracy influence coefficient analysis model according to the accuracy of the traditional Chinese medicine syndrome differentiation model on the syndrome test set data;
calculating an accuracy influence coefficient according to the accuracy influence coefficient analysis model;
the accuracy influence coefficient analysis model is as follows:
in the formula, ACC 0 For the accuracy threshold, ACC is the accuracy,a correction factor for the accuracy influencing coefficient;
the method for acquiring the curve area influence coefficient comprises the following steps:
constructing a curve area influence coefficient analysis model according to the area under the ROC curve of the traditional Chinese medicine syndrome differentiation model on the syndrome test set data;
calculating a curve area influence coefficient according to the curve area influence coefficient analysis model;
the curve area influence coefficient analysis model is as follows:
in AUC 0 The area under the ROC curve threshold, the AUC is the area under the ROC curve, A correction factor for the curve area influence coefficient;
the analysis method of the syndrome positive influence coefficient comprises the following steps:
obtaining the syndrome positive rate and the syndrome positive number of the traditional Chinese medicine syndrome differentiation model on the syndrome test set data;
constructing a syndrome positive influence coefficient analysis model according to the syndrome positive rate and the syndrome positive number;
according to the syndrome positive influence coefficient analysis model, calculating a syndrome positive influence coefficient;
the syndrome positive influence coefficient analysis model is as follows:
in θ 0 For the syndrome positive rate threshold value, theta is the syndrome positive rate, delta theta is the syndrome positive rate set unit value, kappa 1 For the weight ratio of the syndrome positive rate in the syndrome positive influence coefficient, L 0 For the threshold value of the number of syndrome positive cases, L is the number of syndrome positive cases, deltaL is the set unit value of the number of syndrome positive cases, and kappa 2 For the weight ratio of the number of syndrome positive examples in the syndrome positive influence coefficient,is a correction factor for syndrome positive influence coefficient.
9. The method for identifying traditional Chinese medicine symptoms based on intelligent algorithm according to claim 8, wherein the analysis method for the extraction accuracy of the traditional Chinese medicine data is as follows:
acquiring evaluation indexes of a traditional Chinese medicine data extraction model, wherein the evaluation indexes comprise F1 scores;
Constructing a syndrome Chinese medicine data extraction accuracy analysis model according to the evaluation indexes;
calculating the extraction accuracy of the traditional Chinese medicine data according to the analysis model of the extraction accuracy of the traditional Chinese medicine data;
the traditional Chinese medicine data extraction accuracy analysis model is as follows:
wherein F is an evaluation index, F 0 For the evaluation index threshold, deltaF is the unit value of the evaluation index,and extracting correction factors of accuracy rate for the traditional Chinese medicine data.
10. The traditional Chinese medicine syndrome identification device based on the intelligent algorithm is characterized by comprising a data preprocessing module, a syndrome differentiation model construction module, a syndrome differentiation model test module, a syndrome identification ability evaluation module and an auxiliary syndrome differentiation prediction module:
the data preprocessing module is used for: the method comprises the steps of preprocessing traditional Chinese medicine syndrome text data, and dividing the preprocessed traditional Chinese medicine syndrome text data into syndrome training set data and syndrome test set data according to a certain proportion;
the dialectical model construction module is used for: the method is used for constructing a syndrome differentiation model based on syndrome training set data;
the dialectical model test module is as follows: the method is used for testing the syndrome differentiation model based on the syndrome test set data to determine the traditional Chinese medicine syndrome differentiation model;
The syndrome identification capability evaluation module: the comprehensive evaluation index of the syndrome identification capability is used for quantifying the syndrome identification capability of the traditional Chinese medicine syndrome differentiation model;
the auxiliary dialectical prediction module is used for: the intelligent diagnosis interface is used for establishing an intelligent auxiliary diagnosis interface, and the prediction result of the syndrome is output through the intelligent auxiliary diagnosis interface.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410024978.5A CN117766133A (en) | 2024-01-08 | 2024-01-08 | Intelligent algorithm-based traditional Chinese medicine syndrome identification method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410024978.5A CN117766133A (en) | 2024-01-08 | 2024-01-08 | Intelligent algorithm-based traditional Chinese medicine syndrome identification method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117766133A true CN117766133A (en) | 2024-03-26 |
Family
ID=90318392
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410024978.5A Pending CN117766133A (en) | 2024-01-08 | 2024-01-08 | Intelligent algorithm-based traditional Chinese medicine syndrome identification method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117766133A (en) |
-
2024
- 2024-01-08 CN CN202410024978.5A patent/CN117766133A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110491465B (en) | Disease classification coding method, system, device and medium based on deep learning | |
| JP6522161B2 (en) | Medical data analysis method based on deep learning and intelligent analyzer thereof | |
| CN113241135A (en) | Disease risk prediction method and system based on multi-mode fusion | |
| CN111968741B (en) | Deep learning and integrated learning-based diabetes complication high-risk early warning system | |
| CN112100406B (en) | Data processing method, device, equipment and medium | |
| CN110111885B (en) | Attribute prediction method, attribute prediction device, computer equipment and computer readable storage medium | |
| CN111477320B (en) | Treatment effect prediction model construction system, treatment effect prediction system and terminal | |
| CN112489740A (en) | Medical record detection method, training method of related model, related equipment and device | |
| CN107292103A (en) | A kind of prognostic chart picture generation method and device | |
| CN112185558A (en) | Mental health and rehabilitation evaluation method, device and medium based on deep learning | |
| WO2023160264A1 (en) | Medical data processing method and apparatus, and storage medium | |
| CN111477337A (en) | Infectious disease early warning method, system and medium based on individual self-adaptive transmission network | |
| CN109360658A (en) | A kind of the disease pattern method for digging and device of word-based vector model | |
| CN117423423B (en) | Health record integration method, equipment and medium based on convolutional neural network | |
| CN112259232B (en) | VTE risk automatic evaluation system based on deep learning | |
| CN116452592B (en) | Method, device and system for constructing brain vascular disease AI cognitive function evaluation model | |
| CN112071431B (en) | Clinical path automatic generation method and system based on deep learning and knowledge graph | |
| CN113707323A (en) | Disease prediction method, device, equipment and medium based on machine learning | |
| JP7365747B1 (en) | Disease treatment process abnormality identification system based on hierarchical neural network | |
| CN113096756A (en) | Disease evolution classification method and device, electronic equipment and storage medium | |
| CN116884612A (en) | Intelligent analysis method, device, equipment and storage medium for disease risk level | |
| CN116468043A (en) | Nested entity identification method, device, equipment and storage medium | |
| CN110610766A (en) | Apparatus and storage medium for deriving probability of disease based on symptom feature weight | |
| EP4199002A1 (en) | A computer implemented method and a system | |
| CN117766133A (en) | Intelligent algorithm-based traditional Chinese medicine syndrome identification method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |